PageViewTiming: Async or dynamic page details

Browser monitoring's PageViewTiming event sends each data point as a separate event as soon as it is available. Because we do not restrict the timing, you can receive first paint or first interaction data regardless of when it fires. This document describes why and how to use PageViewTiming and its attributes to query data about your site, component loading, and user performance metrics, both from visual and responsiveness standpoints.

Why use PageViewTiming?

If your application uses asynchronous or dynamic pages, you may need additional details about site or component loading. But pages can load content in many different ways, and users control when they interact with that content. This is why some user-centric performance metrics happen outside the standard window onload (page load time) in the browser agent.

For example, users may become impatient and begin clicking as soon as content is on the webpage. Or, they may wait to use the page until long after content is loaded.

The PageViewTiming event provides a more real-time delivery mechanism that does not have a dependency on any other event. The additional metrics can help you understand how users experience your site, both from visual and responsiveness standpoints.

Support for Google's Core Web Vitals

As of agent version 1177 for browser monitoring, we have full support for Google's Core Web Vitals. It requires a Pro or above Browser agent.

Note that the metrics that make up Core Web Vitals evolve over time. The current set focuses on three aspects of the user experience: loading, interactivity, and visual stability. It includes the following metrics and their respective thresholds:

Core Web Vitals metrics include loading, interactivity, and visual stability.

Largest Contentful Paint (LCP): measures loading performance. To provide a good user experience, LCP should occur within 2.5 seconds of when the page first starts loading.
First Input Delay (FID): measures interactivity. To provide a good user experience, pages should have a FID of less than 100 milliseconds.
Cumulative Layout Shift (CLS): measures visual stability. To provide a good user experience, pages should maintain a CLS of less than 0.1.

For each of these metrics, to ensure you're hitting the recommended target for most of your users, a good threshold to measure is the 75th percentile of page loads, segmented across mobile and desktop devices.

To learn more, watch our Nerd Days talk on perceived performance.

Detailed visual, interactivity, and responsiveness metrics

The BrowserInteraction and PageView events end their reporting when they receive the page window load (or window load and AJAX) timing. However, paint and interactivity metrics can happen at any time. PageViewTiming delivers these metrics as a separate event to:

Account for the variability in this timing.
Avoid setting an arbitrary timeout.
Prevent holding BrowserInteraction and PageView events indefinitely.

Additional data	Comments
`firstPaint` and `firstContentfulPaint`	The `firstPaint` and `firstContentfulPaint` attributes already are available with `BrowserInteraction` and `PageView` events. However, they are not always reliably captured before the window onload event fires. Using `PageViewTiming` gives you a way to capture these metrics even if they happen after the original page load time. This gives you a better understanding of the correlation between responsiveness of that load event and the visual rendering of your content.
`largestContentfulPaint`	The `largestContentfulPaint`,metric is available with agent version 1163 or higher. It reports the render time of the largest content element visible in the viewport. Google's research found that looking at when the largest element was rendered was a more accurate way to measure when the main content of a page is loaded and useful. For more information about this metric, including limitations and considerations, see the w3c draft. We also report the cumulative layout shift (CLS) score attribute with LCP. This attribute is reported as `cumulativeLayoutShift`. Largest Contentful Paint is one of three metrics identified by Google as the Core Web Vitals. LCP values up to 2.5 secs are considered "Good," between 2.5-4.0 secs are considered "Needs Improvement," and above 4.0 secs are considered "Poor."
`firstInteraction` and `firstInputDelay`	With the addition of `firstInteraction` and `firstInputDelay`, you can quickly determine the ways that your users are interacting with that visual content. These metrics tell you not only when they interacted, but what type of interaction (mousedown, pointerdown, etc.) and how long it took for them to receive a response from your site. The `firstInputDelay` metric lies in the middle of `FirstContentfulPaint` and Time to Interactive (TTI) metrics. It measures the time between when a first input can be made and when the browser's main thread is able to respond to any interactions. We also report the cumulative layout shift (CLS) score attribute at the moment of the user's first interaction. This attribute is reported as `cumulativeLayoutShift.` First Input Delay is one of three metrics identified by Google as the Core Web Vitals. FID values up to 100 ms are considered "Good," between 100-300 ms are considered "Needs Improvement," and above 300 ms are considered "Poor." For a more detailed explanation, see our browser monitoring release notes.
`cumulativeLayoutShift`	Cumulative Layout Shift (CLS) is available with agent v1177 or higher. CLS is an important, user-centric metric for measuring visual stability because it helps quantify how often users experience unexpected layout shifts. A low CLS helps ensure that the page is delightful. This is one of three metrics identified by Google as the Core Web Vitals. Cumulative Layout Shift is one of three metrics identified by Google as the Core Web Vitals. CLS scores up to 0.1 are considered "Good," between 0.1-0.25 are considered "Needs Improvement," and above 0.25 are considered "Poor."
`timingName`	You can review different types of activities with the `timingName` attribute, such as `firstPaint`, `firstContentfulPaint`, `firstInteraction, largestContentfulPaint, pageHide and windowUnload`. For example, a `PageViewTiming` event may have a `timingName` of `firstPaint` and a `firstPaint` value of `.03`. The event will also include all default attributes included with the standard `BrowserInteraction` and `PageView` events.
`elementId`	This is the `Id`, if specified, of the `largestContentfulPaint` element. This value will only be reported with the LCP metric. This value can be `null`.
`elementSize`	This is the reported size of the `largestContentfulPaint` element. This value will only be reported with the LCP metric.
`pageHide`	The `pageHide` event, available with agent v1177 or higher, is sent when the browser hides the current page in the process of presenting a different page from the session's history. For example, when the user clicks the browser's Back button, the current page receives a `pageHide` event before the previous page is shown. For supporting documentation and browser compatibility for the `pageHide` event, see the MDN Web Docs site. We also report the cumulative layout shift (CLS) score attribute with `pageHide`. This attribute is reported as `cumulativeLayoutShift.`
`windowLoad`	The `windowLoad` event is available with agent v1177 or higher. This is fired when the whole page has loaded, including all dependent resources such as stylesheets and images. For supporting documentation and browser compatibility for the `windowLoad` event, see the MDN Web Docs site. We also report the cumulative layout shift (CLS) score attribute with `windowLoad`. This attribute is reported as `cumulativeLayoutShift.`
`windowUnload`	The `windowUnload` event is available with agent v1163 or higher. This is fired when a document or child resource is being unloaded. For supporting documentation and browser compatibility for the `windowUnload` event, see the MDN Web Docs site. We also report the cumulative layout shift (CLS) score attribute with `windowUnload`. This attribute is reported as `cumulativeLayoutShift`.

Compatibility and requirements

Requirements:

Meets install requirements.
Reporting of this event requires browser agent version 1153 or higher and a Pro or Pro+SPA agent.

Follow our Browser agent release notes to find out when new metrics are released.

These metrics are supported by the following browser versions. For unsupported browsers, no PageViewTiming events will be recorded.

Metrics	Supported browser versions
`cumulativeLayoutShift`	Chrome 79 Metric is elevated to stable; changes in metric definition will be reported in this log. Chrome 77 Metric exposed via API: Cumulative Layout Shift available via Layout Instability API
`firstPaint` `firstContentfulPaint`	Chrome 60 or higher for desktop and mobile (Android webview and Chrome for Android) Opera 47 or higher for desktop Opera 44 or higher for Android mobile Samsung Internet for mobile
`largestContentfulPaint`	Chrome 77 or higher for desktop and mobile
`firstInteraction` `firstInputDelay`	These metrics require the `addEventListener` browser API. This API is available in all modern browsers, including: Apple Safari Google Chrome Microsoft Internet Explorer (IE) versions 9 or higher Mozilla Firefox
`pageHide`	This metric is currently supported by all browsers on desktop and mobile. Compatibility matrix via MDN documentation.
`windowLoad`	This metric is currently supported by all browsers on desktop and mobile. Compatibility matrix via MDN documentation.
`windowUnload`	This metric is currently supported by all browsers on desktop and mobile. Compatibility matrix via MDN documentation.

CumulativeLayoutShift

Cumulative Layout Shift (CLS) is a metric measuring the stability of the content on a webpage. For a complete description, see web.dev/cls.

How is CLS captured in New Relic

Shifts in page layout as reported by the Layout Instability API are aggregated throughout the life of the page and reported as an attribute on all PageViewTiming events, representing the CLS value when that event occurred.

Using this model, users can look at their CLS value at different points in the page's life; for example, CLS values up until the first-time users interact with the page or hide the page.

Approximating other CLS sources

Lighthouse captures CLS value only up to the time when a page is loaded, which is useful in a development or lab environment. You can approximate Lighthouse values by looking at the windowLoad PageViewTiming event.

CrUX report uses values captured over the lifespan of the page, which is useful to analyze worst-case shifts in a RUM environment. You can approximate CrUX values by looking at the CLS attribute on the windowUnload PageViewTiming event. These values will not be exactly the same because of different sample sets and a difference in how values from long-lived web pages are included. The New Relic browser monitoring agent captures CLS when the page unloads, while CrUX collects and updates the metric throughout the lifespan of the page.

How CLS is aggregated

As of July 2021, Google has updated the way CLS values are aggregated. Browser monitoring agent versions v12xx use the method described in Evolving the CLS metric.

Browser monitoring agent v12xx or higher:

Layout shift values are captured in windows. Layout shifts that occurred within 1 second of each other, but no more than 5 seconds between the first and last shift, are part of the same window. A CLS score represents the sum of layout shift values from the window with the highest sum of layout shift values.

Prior to Browser agent v12xx:

A CLS score represents the sum of all layout shifts that occurred up until that point in the page's life.

Query your event data

Here are some sample queries for the event data to help you get started.

Show the 95th percentile of first paint and first contentful paint over a time series:

SELECT FILTER(percentile(firstPaint, 95), where(timingName = 'firstPaint')) as 'fp', FILTER(percentile(firstContentfulPaint, 95), where(timingName = 'firstContentfulPaint')) as 'fcp' FROM PageViewTiming TIMESERIES 1 minute SINCE 1 hour ago

Show the 95th percentile of first input delay over a time series, faceted by transaction name and interaction type:

SELECT percentile(firstInputDelay, 95) as 'fid' FROM PageViewTiming WHERE timingName = 'firstInteraction' TIMESERIES 1 minute FACET browserTransactionName, interactionType SINCE 3 hours ago

Show a histogram of first input delay timings faceted by first interaction time ranges:

FROM PageViewTiming SELECT histogram(firstInputDelay, 1000, 10) SINCE 3 hours ago WHERE timingName = 'firstInteraction' FACET CASES (WHERE firstInteraction < 1, WHERE firstInteraction >= 1 AND firstInteraction < 5, WHERE firstInteraction >= 5)